PR: Drop Python 2

[dalthviz]

Fixes #250

[avalentino]

Python 2.7 is also mentioned in a comment in qtpy/tests/test_qtpositioning.py.

[CAM-Gerlach]

Very thorough job, @dalthviz ! I ran a bunch of regexes I use to detect possible Py <3.6 blocks and code, as well as manually looked through both your changes and the codebase, and everything looked great. I also ran pyupgrade on the repo with pyupgrade --py36-plus setup.py qtpy/*.py qtpy/tests/*.py qtpy/_patch/*.py, which did spot and autofix a few remaining issues that allowed eliminating/simplifying a further >100 lines of code:

• Legacy absolute_import future imports that were missed in the tests
• Calls to the mock third-party backport instead of the built-in unittest.mock, which allowed eliminating it as an extra test dependency
• Redundant coding: utf-8 declarations (since the default source file encoding on Python 3+ is UTF-8)
• A few uses of legacy ordered format() calls that could be safety rewritten as much simpler f-strings
• A few other small odds and ends (a legacy super() call, a legacy class declaration, and importing and using io.open() instead of the now-identical and more concise/standard builtin open()

The results of the run plus the accompanying manual tweaks (removing the now-unused import io and dropping mock from the test scripts) are on the fix_issue_250 branch on my fork. Assuming you agree with the changes, feel free to pull that and push here, or I can do so directly if that's easier.

[dalthviz]

Hi @CAM-Gerlach thank you for checking and futher helping with this! I will check the new changes and pull them here 👍

[CAM-Gerlach]

I assume the CI issues are a result of rebasing on #225 , as I didn't see anything obviously connected to my changes.

However, just in case you didn't already see it, in test-pyside6.sh a dead elif block for Python 2.7 got added, aside from a couple of trivial things (coding marker in QtWebEngine.py, old-style format in pytest_report_header in conftest.py),

[CAM-Gerlach]

LGTM @dalthviz ; thanks as always for all your hard work

[dalthviz]

Hi @CAM-Gerlach , could you do a new review here to check if after the last rebase everything is still good? Thanks!

Edit: Also, if is good go ahead and merge it 👍

[CAM-Gerlach]

Sure, in work 👍

[CAM-Gerlach]

Thanks so much for all your work here, @dalthviz . I reviewed all the code changes, regexed for any remaining vestiges of Qt4/Python 2, and re-ran pyupgrade, and everything looks good aside from two issues. Since GitHub wouldn't let me make one of them as a suggestion, I just made them as separate commits (for ease of cherry-picking/dropping/fixuping etc) on my branch, like before, so if you like, you can easily just pull that in again and cherry-pick/squash/fixup/etc. to your liking if you agree with the changes, or if you decline the one I couldn't make as a suggestion, feel free to just apply the suggestion here.

Specifically, in addition to the comment about mock, not sure if it happened in the rebase, but most of lines 89-102 in _qfiledialog_wrapper() in compat.py appear to be dead code due to being only for Qt <=4.6, and can be replaced by just:

    result = func(parent, caption, basedir, filters, selectedfilter, options)
    if sys.platform == "win32":
        # On Windows platforms: restore standard outputs
        sys.stdout, sys.stderr = _temp1, _temp2

(See a1380d7)

Out of date after rebase

[CAM-Gerlach]

Oops, forgot to force push to the branch on my fork, sorry. Its up to date now; specifically a1380d7 is the commit with the Qt4 dead code cleanup that might have been missed during the rebase.

[CAM-Gerlach]

LGTM, thanks so much @dalthviz